We have measured bone mineral density (BMD) using dual X-ray absorptiometry (DXA) of the spine and hip, spinal quantitative computed tomography (QCTspi), and peripheral radial quantitative computed tomography (pQCTrad) in 334 spine and 51 hip fracture patients. The standardized hip and spine BMD for each patient was calculated and compared with the combined reference ranges published previously, each densitometer having been cross- calibrated with the prototype European Spine Phantom (ESPp) or the European Forearm Phantom (EFP). Male and female fracture cases had similar BMD values after adjusting for body size, where appropriate. This suggests that the relationship between bone density (mass per unit volume) and fracture risk is similar between men and women. However, compared with age-matched controls, mean decreases in BMD ranged from 0.78 SD units (women with hip fracture, DXAspi) to 2.57 SD units (men with spine fractures, QCTspi). The proportion of spine and hip fracture patients falling below the cutoff for osteoporosis (T-score <-2.5 SD) proposed by the World Health Organization (WHO) study group varied according to different BMD measurement procedures (range 18- 94%). This finding suggests that the WHO definition requires different thresholds when used with non-DXA BMD measurement techniques. Receiver operator characteristic (ROC) analysis was used to compare measurement techniques for their ability to discriminate between cases and controls. Among DXA sites, the proximal femur was preferred when evaluating generalized bone loss, particularly in elderly people. An additional spinal BMD measurement may add clinical value if spine fracture risk assessment has a high priority. Both axial and peripheral QCT techniques performed comparably to DXA in spinal osteoporosis, so investigators and clinicians may use any of the three technologies with similar degrees of confidence for the diagnosis of generalized or site-specific bone loss providing straightforward clinical guidelines are followed.